Gemma 3 12B: how much VRAM?
In Q4_K_M with 8k tokens of context, Gemma 3 12B needs about 10 GB of memory. The cheapest option: RTX 3060 12 GB (€1,180 for the complete machine).
Model sheet
ollama run gemma3:12bWeights on Hugging Face: google/gemma-3-12b-it
VRAM by precision and context
| Precision | 4k | 8k | 32k | 128k |
|---|---|---|---|---|
| FP16 | 26 GB | 27 GB | 32 GB | 51 GB |
| FP8 | 14 GB | 15 GB | 19 GB | 39 GB |
| Q8_0 | 15 GB | 15 GB | 20 GB | 39 GB |
| Q6_K | 12 GB | 12 GB | 17 GB | 36 GB |
| Q4_K_M | 9 GB | 10 GB | 14 GB | 34 GB |
| Q3_K_M | 7 GB | 8 GB | 13 GB | 32 GB |
One user. Each simultaneous user adds their own context cache.
Frequently asked questions
How much VRAM does Gemma 3 12B need?
With an 8k-token context and one user, Gemma 3 12B needs about 10 GB in Q4, 15 GB in Q8 and 27 GB in FP16.
What is the cheapest hardware to run Gemma 3 12B?
In Q4, the cheapest option in our catalogue to run Gemma 3 12B is RTX 3060 12 GB, about €1,180 for the complete machine.
How fast does Gemma 3 12B run?
Expect around 27 tokens per second for one user on RTX 3060 12 GB in Q4 (estimate based on memory bandwidth).
On which hardware?
| Hardware | Cards | Speed | Machine |
|---|---|---|---|
| RTX 3060 12 GB | 1 is enough | ~27 tok/s | €1,180 |
| RTX 4060 Ti 16 GB | 1 is enough | ~22 tok/s | €1,350 |
| Radeon RX 7800 XT 16 GB | 1 is enough | ~47 tok/s | €1,400 |
| RTX 4070 12 GB | 1 is enough | ~38 tok/s | €1,450 |
| RTX 3090 24 GB (used) | 1 is enough | ~70 tok/s | €1,650 |
| RTX 4070 Ti SUPER 16 GB | 1 is enough | ~50 tok/s | €1,750 |
| Radeon RX 7900 XTX 24 GB | 1 is enough | ~72 tok/s | €1,850 |
| RTX 4080 SUPER 16 GB | 1 is enough | ~55 tok/s | €1,950 |
| RTX 5080 16 GB | 1 is enough | ~72 tok/s | €2,100 |
| Mini-PC Ryzen AI Max+ 395 128 GB | 1 is enough | ~19 tok/s | €2,300 |
| RTX 4090 24 GB | 1 is enough | ~75 tok/s | €2,900 |
| RTX 5090 32 GB | 1 is enough | ~134 tok/s | €3,500 |
| NVIDIA DGX Spark 128 GB | 1 is enough | ~20 tok/s | €4,000 |
| Mac Studio M4 Max 128 GB | 1 is enough | ~41 tok/s | €4,400 |
| RTX 6000 Ada 48 GB | 1 is enough | ~72 tok/s | €7,900 |
| RTX PRO 6000 Blackwell 96 GB | 1 is enough | ~134 tok/s | €9,900 |
| Mac Studio M3 Ultra 512 GB | 1 is enough | ~61 tok/s | €11,500 |
| L40S 48 GB | 1 is enough | ~65 tok/s | €13,500 |
| A100 80 GB | 1 is enough | ~153 tok/s | €17,000 |
| H100 80 GB | 1 is enough | ~251 tok/s | €34,000 |
| H200 141 GB | 1 is enough | ~359 tok/s | €39,000 |
| B200 192 GB | 1 is enough | ~599 tok/s | €48,000 |
Estimated complete machine: cards + host PC (€900 for 1–2 consumer cards, €2,800 beyond, a €6,000 server for datacenter cards). Macs and mini-PCs are sold complete.
Similar models
From: RTX 3060 12 GB · €1,180
From: RTX 3060 12 GB · €1,180
From: RTX 3060 12 GB · €1,180
From: RTX 3060 12 GB · €1,180